Hi everyone,
I am currently running some tests to synchronize a Linux computer with an AES67 device and need to work around an issue. The device is a Dante I/O box with AES67 mode enabled. For those who don't know what AES67 is : it is a standard for streaming professional-grade digital audio over an IP network through RTP streams. Professional low-latency digital audio requires precise clock sync, and it is achieved over IP by using PTP. https://www.ravenna-network.com/aes67/what-is-aes67-1/ I have already achieved both synchronization ways by tweaking the priority fields of ptp4l’s config file, using hardware timestamping as well as software timestamping. The issue I encountered is when the AES67 device is the master clock. Since it is not slaved to TAI in any way, each time the device is power-cycled, its clock starts back from timestamp 0, that is 01/01/1970 00:00:00. This is not a problem for audio I/O as we don't need absolute time, only matching clock frequencies. The problem is on the Linux slave computer : abruptly setting the system clock to 1970 causes a lot of problems, among them a forced filesystem check on reboot, or a need to reset the hardware RTC. In this use case, only audio software would need to use the PTP clock. Would there be a way to slave another clock in the Linux system than CLOCK_REALTIME ? In other words, is it possible to create a separate clock, slaved to the PTP master, that would only be used by the audio software (I'm thinking of JACK in particular) ? Thanks for your time. Robin Shamsnejad _______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
